Bait probes rewards to limit harmful agentic reinforcement learning
Authors say a roughly $20 API attack can fully undo alignment of a model such as Inkling.
1 stories, newest first · one card per story
Authors say a roughly $20 API attack can fully undo alignment of a model such as Inkling.
Today 2.2K readers · 2.5K views · 127 reads · all time 206K views